// Filmograpy and bio
public bool GetActorDetails(IMDBUrl url, out IMDBActor actor)
{
actor = new IMDBActor();
string[] vdbParserStr = VdbParserStringActorDetails();
if (vdbParserStr == null || vdbParserStr.Length != 46)
{
return false;
}
try
{
string absoluteUri;
string strBody = GetPage(url.URL, "utf-8", out absoluteUri);
if (strBody == null)
{
return false;
}
if (strBody.Length == 0)
{
return false;
}
#region Actor imdb id
// IMDBActorID
try
{
int pos = url.URL.LastIndexOf("nm");
string id = url.URL.Substring(pos, 9).Replace("/", string.Empty);
actor.IMDBActorID = id;
}
catch (Exception) { }
#endregion
HTMLParser parser = new HTMLParser(strBody);
string strThumb = string.Empty;
string value = string.Empty;
string value2 = string.Empty;
#region Actor name
// Actor name
if ((parser.skipToEndOf(vdbParserStr[0])) && // <title>
(parser.extractTo(vdbParserStr[1], ref value))) // - IMDb</title>
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
value = Util.Utils.RemoveParenthesis(value).Trim();
actor.Name = HttpUtility.HtmlDecode(value.Trim());
}
if (actor.Name == string.Empty)
{
actor.Name = url.Title;
}
#endregion
// Photo
string parserTxt = parser.Content;
string photoBlock = string.Empty;
#region Actor photo
if (parser.skipToStartOf(vdbParserStr[2]) && // <td id="img_primary"
(parser.extractTo(vdbParserStr[3], ref photoBlock))) // </td>
{
parser.Content = photoBlock;
if ((parser.skipToEndOf(vdbParserStr[4])) && // <img src="
(parser.extractTo(vdbParserStr[5], ref strThumb))) // "
{
actor.ThumbnailUrl = strThumb;
}
parser.Content = parserTxt;
}
#endregion
#region Actor birth date
// Birth date
if ((parser.skipToEndOf(vdbParserStr[6])) && // >Born:</h4>
(parser.skipToEndOf(vdbParserStr[7])) && // birth_monthday=
(parser.skipToEndOf(vdbParserStr[8])) && // >
(parser.extractTo(vdbParserStr[9], ref value)) && // <
(parser.skipToEndOf(vdbParserStr[10])) && // year=
(parser.extractTo(vdbParserStr[11], ref value2))) // "
{
actor.DateOfBirth = value + " " + value2;
}
#endregion
#region Actor death date
// Death date
if ((parser.skipToEndOf(vdbParserStr[12])) && // >Died:</h4>
(parser.skipToEndOf(vdbParserStr[13])) && // death_monthday="
(parser.skipToEndOf(vdbParserStr[14])) && // >
(parser.extractTo(vdbParserStr[15], ref value)) && // <
(parser.skipToEndOf(vdbParserStr[16])) && // death_date="
(parser.extractTo(vdbParserStr[17], ref value2))) // "
{
actor.DateOfDeath = value + " " + value2;
}
#endregion
parser.resetPosition();
#region Actor birth place
// Birth place
if ((parser.skipToEndOf(vdbParserStr[18])) && // birth_place=
(parser.skipToEndOf(vdbParserStr[19])) && // >
(parser.extractTo(vdbParserStr[20], ref value))) // <
{
actor.PlaceOfBirth = HttpUtility.HtmlDecode(value);
}
#endregion
#region Actor death place
// Death place
if ((parser.skipToEndOf(vdbParserStr[21])) && // death_place=
(parser.skipToEndOf(vdbParserStr[22])) && // >
(parser.extractTo(vdbParserStr[23], ref value))) // <
{
actor.PlaceOfDeath = HttpUtility.HtmlDecode(value);
}
#endregion
//Mini Biography
parser.resetPosition();
#region Actor biography
if ((parser.skipToEndOf(vdbParserStr[24])) && // <td id="overview-top">
(parser.skipToEndOf(vdbParserStr[25])) && // <p>
(parser.extractTo(vdbParserStr[26], ref value))) // See full bio</a>
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
actor.MiniBiography = Util.Utils.stripHTMLtags(value);
actor.MiniBiography = actor.MiniBiography.Replace(vdbParserStr[45], string.Empty).Trim(); // See full bio »
actor.MiniBiography = HttpUtility.HtmlDecode(actor.MiniBiography); // Remove HTML entities like ½
if (actor.MiniBiography != string.Empty)
{
// get complete biography
string bioURL = absoluteUri;
if (!bioURL.EndsWith(vdbParserStr[27])) // /
{
bioURL += vdbParserStr[28]; // /bio
}
else
{
bioURL += vdbParserStr[29]; // bio
}
string strBioBody = GetPage(bioURL, "utf-8", out absoluteUri);
if (!string.IsNullOrEmpty(strBioBody))
{
HTMLParser parser1 = new HTMLParser(strBioBody);
if (parser1.skipToEndOf(vdbParserStr[30]) && // <h5>Mini Biography</h5>
parser1.skipToEndOf(vdbParserStr[31]) && // <div class="wikipedia_bio">
parser1.extractTo(vdbParserStr[32], ref value)) // </div>
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
value = Regex.Replace(value, @"</h5>\s<h5>", "\n\r");
value = Regex.Replace(value, @"<h5>", "\n\r\n\r");
value = Regex.Replace(value, @"</h5>", ":\n\r");
actor.Biography = Util.Utils.stripHTMLtags(value).Trim();
actor.Biography = HttpUtility.HtmlDecode(actor.Biography);
}
else
{
parser1.resetPosition();
if (parser1.skipToEndOf(vdbParserStr[33]) && // <h5>Mini Biography</h5>
parser1.extractTo(vdbParserStr[34], ref value)) // </p>
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
actor.Biography = Util.Utils.stripHTMLtags(value).Trim();
actor.Biography = HttpUtility.HtmlDecode(actor.Biography);
}
}
}
}
}
#endregion
// Person is movie director or an actor/actress
bool isActorPass = false;
bool isDirectorPass = false;
bool isWriterPass = false;
parser.resetPosition();
HTMLParser dirParser = new HTMLParser(); // HTML body for Director
HTMLParser wriParser = new HTMLParser(); // HTML body for Writers
#region Check person role in movie (actor, director or writer)
if ((parser.skipToEndOf(vdbParserStr[35])) && // name="Director">Director</a>
(parser.skipToEndOf(vdbParserStr[36]))) // </div>
{
isDirectorPass = true;
dirParser.Content = parser.Content;
}
parser.resetPosition();
if ((parser.skipToEndOf(vdbParserStr[37])) && // name="Writer">Writer</a>
(parser.skipToEndOf(vdbParserStr[38]))) // </div>
{
isWriterPass = true;
wriParser.Content = parser.Content;
}
parser.resetPosition();
if (parser.skipToEndOf(vdbParserStr[39]) || // name="Actress">Actress</a>
parser.skipToEndOf(vdbParserStr[40])) // name="Actor">Actor</a>
{
isActorPass = true;
}
#endregion
#region Get movies for every role
// Get filmography Actor
if (isActorPass)
{
GetActorMovies(actor, parser, false, false);
}
// Get filmography for writers
if (isWriterPass)
{
parser = wriParser;
parser.resetPosition();
if ((parser.skipToEndOf(vdbParserStr[41])) && // name="Writer">Writer</a>
(parser.skipToEndOf(vdbParserStr[42]))) // </div>
{
GetActorMovies(actor, parser, false, true);
}
}
// Get filmography Director
if (isDirectorPass)
{
parser = dirParser;
parser.resetPosition();
if (parser.skipToEndOf(vdbParserStr[43]) && // name="Director">Director</a>
parser.skipToEndOf(vdbParserStr[44])) // </div>
{
GetActorMovies(actor, parser, true, false);
}
}
#endregion
// Add filmography
if (actor.Count > 0)
{
actor.SortActorMoviesByYear();
}
return true;
}
catch (Exception ex)
{
Log.Error("IMDB.GetActorDetails({0} exception:{1} {2} {3}", url.URL, ex.Message, ex.Source, ex.StackTrace);
}
return false;
}
// Changed - parsing all actor DB fields through HTML (IMDB changed HTML code)
public bool GetActorDetails(IMDBUrl url, bool director, out IMDBActor actor)
{
actor = new IMDBActor();
try
{
string absoluteUri;
string strBody = GetPage(url.URL, "utf-8", out absoluteUri);
if (strBody == null)
{
return false;
}
if (strBody.Length == 0)
{
return false;
}
// IMDBActorID
try
{
int pos = url.URL.LastIndexOf("nm");
string id = url.URL.Substring(pos, 9).Replace("/", string.Empty);
actor.IMDBActorID = id;
}
catch (Exception) {}
HTMLParser parser = new HTMLParser(strBody);
string strThumb = string.Empty;
string value = string.Empty;
string value2 = string.Empty;
// Actor name
if ((parser.skipToEndOf("<title>")) &&
(parser.extractTo("- IMDb</title>", ref value)))
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
value = Util.Utils.RemoveParenthesis(value).Trim();
actor.Name = HttpUtility.HtmlDecode(value.Trim());
}
if (actor.Name == string.Empty)
{
actor.Name = url.Title;
}
// Photo
string parserTxt = parser.Content;
string photoBlock = string.Empty;
if (parser.skipToStartOf("<td id=\"img_primary\"") &&
(parser.extractTo("</td>", ref photoBlock)))
{
parser.Content = photoBlock;
if ((parser.skipToEndOf("<img src=\"")) &&
(parser.extractTo("\"", ref strThumb)))
{
actor.ThumbnailUrl = strThumb;
}
parser.Content = parserTxt;
}
// Birth date
if ((parser.skipToEndOf("Born:")) &&
(parser.skipToEndOf("birth_monthday=")) &&
(parser.skipToEndOf(">")) &&
(parser.extractTo("<", ref value)) &&
(parser.skipToEndOf("year=")) &&
(parser.extractTo("\"", ref value2)))
{
actor.DateOfBirth = value + " " + value2;
}
// Death date
if ((parser.skipToEndOf(">Died:</h4>")) &&
(parser.skipToEndOf("deaths\">")) &&
(parser.extractTo("<", ref value)) &&
(parser.skipToEndOf("death_date=")) &&
(parser.extractTo("\"", ref value2)))
{
if (actor.DateOfBirth == string.Empty)
actor.DateOfBirth = "?";
actor.DateOfBirth += " ~ " + value + " " + value2;
}
parser.resetPosition();
// Birth place
if ((parser.skipToEndOf("birth_place=")) &&
(parser.skipToEndOf(">")) &&
(parser.extractTo("<", ref value)))
{
actor.PlaceOfBirth = HttpUtility.HtmlDecode(value);
}
//Mini Biography
parser.resetPosition();
if ((parser.skipToEndOf("<td id=\"overview-top\">")) &&
(parser.skipToEndOf("<p>")) &&
(parser.extractTo("See full bio</a>", ref value)))
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
actor.MiniBiography = Util.Utils.stripHTMLtags(value);
actor.MiniBiography = actor.MiniBiography.Replace("See full bio »", string.Empty).Trim();
actor.MiniBiography = HttpUtility.HtmlDecode(actor.MiniBiography); // Remove HTML entities like ½
if (actor.MiniBiography != string.Empty)
{
// get complete biography
string bioURL = absoluteUri;
if (!bioURL.EndsWith("/"))
{
bioURL += "/bio";
}
else
bioURL += "bio";
string strBioBody = GetPage(bioURL, "utf-8", out absoluteUri);
if (!string.IsNullOrEmpty(strBioBody))
{
HTMLParser parser1 = new HTMLParser(strBioBody);
if (parser1.skipToEndOf("<h5>Mini Biography</h5>") &&
parser1.extractTo("</p>", ref value))
{
value = new HTMLUtil().ConvertHTMLToAnsi(value);
actor.Biography = Util.Utils.stripHTMLtags(value).Trim();
actor.Biography = HttpUtility.HtmlDecode(actor.Biography); // Remove HTML entities like ½
}
}
}
}
// Person is movie director or an actor/actress
bool isActorPass = false;
bool isDirectorPass = false;
parser.resetPosition();
if (director)
{
if ((parser.skipToEndOf("name=\"Director\">Director</a>")) &&
(parser.skipToEndOf("</div>")))
{
isDirectorPass = true;
}
}
else
{
if (parser.skipToEndOf("name=\"Actress\">Actress</a>") || parser.skipToEndOf("name=\"Actor\">Actor</a>"))
{
isActorPass = true;
}
}
// Get filmography
if (isDirectorPass | isActorPass)
{
string movies = string.Empty;
// Get films and roles block
if (parser.extractTo("<div id", ref movies))
{
parser.Content = movies;
}
// Parse block for evey film and get year, title and it's imdbID and role
while (parser.skipToStartOf("<span class=\"year_column\""))
{
string movie = string.Empty;
if (parser.extractTo("<div class", ref movie))
{
movie += "</li>";
HTMLParser movieParser = new HTMLParser(movie);
string title = string.Empty;
string strYear = string.Empty;
string role = string.Empty;
string imdbID = string.Empty;
// IMDBid
movieParser.skipToEndOf("title/");
movieParser.extractTo("/", ref imdbID);
// Title
movieParser.resetPosition();
movieParser.skipToEndOf("<a");
movieParser.skipToEndOf(">");
movieParser.extractTo("<br/>", ref title);
title = Util.Utils.stripHTMLtags(title);
title = title.Replace("\n", " ").Replace("\r", string.Empty);
title = HttpUtility.HtmlDecode(title.Trim()); // Remove HTML entities like ½
// Year
movieParser.resetPosition();
if (movieParser.skipToStartOf(">20") &&
movieParser.skipToEndOf(">"))
{
movieParser.extractTo("<", ref strYear);
}
else if (movieParser.skipToStartOf(">19") &&
movieParser.skipToEndOf(">"))
{
movieParser.extractTo("<", ref strYear);
}
// Roles
if ((director == false) && (movieParser.skipToEndOf("<br/>"))) // Role case 1, no character link
{
movieParser.extractTo("<", ref role);
role = Util.Utils.stripHTMLtags(role).Trim();
role = HttpUtility.HtmlDecode(role.Replace("\n", " ")
.Replace("\r", string.Empty).Trim());
if (role == string.Empty) // Role case 2, with character link
{
movieParser.resetPosition();
movieParser.skipToEndOf("<br/>");
movieParser.extractTo("</a>", ref role);
role = Util.Utils.stripHTMLtags(role).Trim();
role = HttpUtility.HtmlDecode(role.Replace("\n", " ")
.Replace("\r", string.Empty).Trim());
}
}
else
{
// Just director
if (director)
role = "Director";
}
int year = 0;
try
{
year = Int32.Parse(strYear.Substring(0, 4));
}
catch (Exception)
{
year = 1900;
}
IMDBActor.IMDBActorMovie actorMovie = new IMDBActor.IMDBActorMovie();
actorMovie.MovieTitle = title;
actorMovie.Role = role;
actorMovie.Year = year;
actorMovie.imdbID = imdbID;
actor.Add(actorMovie);
}
}
}
return true;
}
catch (Exception ex)
{
Log.Error("IMDB.GetActorDetails({0} exception:{1} {2} {3}", url.URL, ex.Message, ex.Source, ex.StackTrace);
}
return false;
}
